# Reply draft: measuring GEO for a B2B manufacturer

Disclosure: I work on MadeInRed, a GEO content tool. I have not run a manufacturing campaign for your client, so this is a proposed way to investigate, not a result claim.

Since you've already improved the site and tracking, I'd take one product family and ask the sales engineer which questions actually decide whether an enquiry is worth continuing. For a custom component, “what do you need from my drawing to quote?” is more useful than a broad “best manufacturer” prompt.

Save a small fixed set of those questions, the AI interface/search setting, language, date, full answer and exact source URLs. Record whether the company is named separately from whether its own page is linked. Keep the same core set for later comparison and expect answers to vary. Add identifiable AI referrals and qualified enquiries from your existing tracking as separate rows; a citation count isn't a lead count.

For the content work, pick one cited answer and check the actual buyer decision. Can the engineer find the relevant material, conditions, exclusions and next-step information on your client's page? If not, fix that answer with approved specifications and a technical review. Don't add a supposed tolerance, certification or lead time just to make the page look complete. Useful specification text and a clear quotation checklist may help more than another generic blog post.

I'd postpone a large tool purchase until that small exercise shows what you need to automate: recurring observation, reviewing citations, or producing content. For Google's AI search features, the official guidance says normal SEO foundations still apply and their traffic is included in overall Web search performance; that isn't a complete report of activity in every AI product.

The first check I'd make with your client is therefore: which one application question does the current site still fail to answer, and who can approve the factual answer?